BLOG

WHAT IS UNCANNY VALLEY?

14.10.19

You may have heard of the uncanny valley effect before. It’s a memorable name, that’s for sure, but what does it pertain to? Take a look at the photograph below...

What strikes you about this image? Her clothes? Her hair colour? The fact that “she” is not a “her” at all, but a robot designed to look like a human being – and with seriously creepy and unsettling results!

That uneasy feeling is central to the notion of uncanny valley, first hypothesised by Masahiro Mori - a Japanese roboticist - way back in 1970. Nearly 50 years ago, Mori mused in his essay “Bukimi No Tani” that as robots become more humanlike, they appear more familiar until a point is reached at which subtle imperfections of appearance make them look eerie. The “valley” represents the way a viewer's level of comfort drops as a simulation approaches, but does not reach, verisimilitude, leading Mori to opine that overly lifelike appearances should be avoided by robot builders.

But it’s not just robots that provoke this negative reaction.

The rise of CGI and 3D animations in movies and video games has also had a discernible impact. Some designers choose to exploit the effect to help heighten or prompt a certain emotional response from users; others actively try to avoid it. 

What’s of particular interest, however, is how differently audiences can react when the effect is – consciously or not - amplified on the big screen. 

 

The Polar Express is a prime example. A 2004 animated adventure led by Tom Hanks received mixed reviews from critics and audiences alike.  

 

Ed Park of Village Voice gave the film a Metascore of 50 out of 100 on its release, lamenting that:

 

“when it comes to the ‘humans’, the atmosphere collapses. Unnervingly smooth, mouths moving in strange, even frightening formations, the Polar people are the least convincing things on-screen, glaring imposters amid the otherwise painstakingly rendered scenery.”

Similarly, Maitland McDonagh from TV Guide Magazine slammed the movie: 

 

“The film’s characters, computer-animated over motion-capture footage of flesh-and-blood performers, are as blank-eyed and rubbery-looking as moving mannequins… the stuff of nightmares, not dreams.”

By contrast, both The Hollywood Reporter and San Francisco Chronicle critics awarded the film the maximum Metascore of 100. “An enchanting, beautiful and brilliantly imagined film”, enthused the latter.

If nothing else, the film demonstrates how pivotally important realism can be to viewers. For those overwhelmed by the uncanny valley effect, their afternoon outing to the cinema is tainted – perhaps even ruined. After all, it is not unfeasible to expect that creepy, unsettling characters are the most enduring ones.

 

For those affected by uncanny valley, the discomfort felt will surely always win out over the positives the film had to offer. It doesn’t matter if Tom Hanks gave the vocal performance of his life as the Conductor (although we all know that accolade is reserved for Woody in Toy Story), if the animations made you feel uneasy it is very unlikely that you’ll endure watching it again. 

Interestingly, research into the phenomena didn’t begin in earnest until as recently as 2005, when Mori’s paper was translated into English by Karl MacDorman and Takashi Minato. A mere 35 academic papers were based on the topic in 2004, compared to more than 500 in 2005 alone.

What robotists, philosophers and psychologists fail to agree on, however, is whether uncanny valley exists at all. Researching such a subjective matter is notoriously tricky, and it is difficult to determine where the boundaries lie. 

Research by Christine Looser and Thalia Wheatley found a valley at the point where inanimate mannequin faces were morphed into human faces and started to look alive. 

Similarly, a study by Kurt Gray and Daniel Wegner deemed that robots are only unnerving when they are judged to be able to sense and experience things, akin to humans. 

On this basis, it’s surely only a matter of time before uncanny valley becomes a thing of the past. The androids of the future will likely be so sophisticated in their appearance, mannerisms and actions that it will be impossible to tell them apart from their human counterparts. 

Who knows, perhaps we’ll rely on Voigt-Kampff tests in the future, à la Do Androids Dream of Electric Sheep? These empathy tests – first referenced in Philip K. Dick’s 1968 novel and later in the 1982 film Blade Runner - are designed to differentiate humans from androids. By measuring brain activity and eye movement, the tests judge the responses of “replicants” to emotional triggers - such as the harming of animals - that would rouse empathetic feelings in human beings.

 

Whilst these fictional tests proved to be unreliable - in a similar vein to the fallibility of the Turing Test - it is a scary thought that there may be no effective method of determining who or what is “real” in years to come. If we know what “real” means at all…

  • LinkedIn
  • Twitter
  • YouTube